Project Summary
Acute brain injuries (e.g., stroke and trauma) are among the leading causes of death and disability in the US.
Patients with acute brain injuries are at high risk for seizures and seizure-like activity as detected by
electroencephalography (EEG). This pathologic epileptiform activity further increases the risk of death and
disability in patients with brain injuries. Anti-seizure medications are frequently used to suppress this activity,
which while potentially efficacious in suppressing the activity also have significant side effects e.g., cognitive and
functional impairment. Despite the significance (i.e., frequency and severity of the injuries and the difficulty
balancing adverse effects of seizures and seizure-like activity with those from the treatment), there is limited
evidence to guide clinical care. The reasons for the gap in knowledge are multifactorial including the complex
and heterogeneous nature of both acute brain injury and seizure-like activity, the labor-intensive task of reviewing
EEG data, and the potential difficulty recruiting and following up with this critically ill patient population which has
historically had limited involvement in clinical trials. To address these questions and overcome these barriers,
we have developed a large multi-center EEG registry and tools for processing and analyzing the EEG data and
have completed preliminary studies to pave the way for the proposed work. Specifically, we will link our EEG
registry data with electronic records, a stroke registry, administrative data and state death records, to achieve
two aims: 1) Aim 1: Describe the natural history of epileptiform activity and its response to anti-seizure
medications using pharmacokinetic and pharmacodynamics modeling, and characterize how this response
varies with injury type and severity; and Aim 2: Examine the impact of EEG-guided anti-seizure treatment on
neurologic outcome (death and disability) using both target trial emulation and reinforcement learning. Our
dataset will be one of the largest linked EEG datasets extant. This comparative effectiveness study will directly
inform and improve the care of acute brain injury patients with seizure-like activity detected on EEG, future
experimental and observational studies, and neurocritical care management practices.